Processing of Amateur Esthetic Digital Astronomical Images
Since digital imaging has become popular in amateur astronomy, the subject of editing digital images has received more attention. Most people are aware that images can be easily altered to virtually any degree by the use of image editing programs such as Adobe Photoshop and others.
So that users of my images might feel more comfortable with my images, I would like to provide a brief description of the most common types of processing that I normally apply to my images and my rationale for doing so. I will not attempt to go into detail as to how this is done (that would probably take a hundred pages) but rather what the effects of the processing are so that the user understands the need for such processing and can judge for themselves how it changes the image. Please also realize that these are not the only techniques I use, but they are a good representative sample of both the type and degree of image processing that I normally perform.
"Linear" vs. "Non-Linear" scaling
Linear scaling is what is used by a large majority of conventional photos of normal earthbound objects. This is a "what you see is what you get" approach. The brightness of a part of the image as seen on a monitor is more or less the same as the brightness of that part of the object as seen in nature. They are thereby said to have a "linear" relationship.
This presents a huge problem for astronomical objects, however. The range of brightness in these objects is often MUCH greater than either the eye can see or a monitor or photo can display. If linear scaling is used for such images, you cannot show detail in both the bright and the dim areas of the image - you would either show the bright areas as all white or the dark areas as all black. A common example is the bright core of many galaxies. In order to show detail in these areas without losing the detail in the dim spiral arms, we must apply something other than linear scaling.
Another process commonly applied is a sharpening process. There are many ways to do this but the essence of the process is that small scale differences are emphasized to reveal more detail in the object. Overdoing this can result in artifact and/or excessive noise, so I try to keep this process to a minimum. The example below shows the effect on an M66 image of a non-linear scaling plus sharpening algorithm commonly used in amateur astronomy (Digital Development Process, or DDP). Exactly the same data was used in both images. I think it is clear that not only is the second image more esthetic, it is also more informative.
Noise is the enemy of astronomical imaging. Most objects are very dim and require a lot of exposure time to produce a clear image. There are, however, limits to the amount of time one can expect to spend taking a photo of an object as well as noise contributions from the atmosphere, the camera, and other sources. Since these can be minimized but never eliminated, it is common to apply processing designed to reduce the effect of noise on an image. Noise is also more common and more noticeable in the dimmer portions of an image so noise reduction may be applied more in those areas as this minimizes the modification of the image. In example shown below, the color channel has been slightly blurred to reduce color noise in an area where little or no overall color was present. As you can see, the subject of the image (the galaxy) was unaffected by this processing.
Finally there are artifacts in most images. These I define as "stuff that isn't really there". One example of an artifact is a vertical streak around bright stars referred to as "blooming". This is a camera electronics problem caused by too much light in a small area. Another example is the radial spiking seen around many bright stars taken with telescopes that have vanes to hold their secondary mirror. These "diffraction spikes" are caused by the interaction of light with these vanes. Yet another kind of artifact can be introduced by processing of the image. These are also most common around bright stars and can be a light "fuzz", a dark "halo", or sometimes both. Such artifacts are commonly corrected manually in a program such as Photoshop by importing a small area of adjacent image to replace the defect. See examples of all of these artifacts and their corrections below.
* Star size has been reduced from the large, blasted-out, bloomed star. Differential scaling and blurring have also been done to improve galaxy rendition and reduce background noise outside the galaxy (as in previous example).
If you has any questions as to what was processing was performed on a particular image, I would be happy to tell you. Please be aware that with older images I may not remember all of the processing details.
For a brief discussion of my philosophy on image processing you can go to my image processing ethics page.
- Bill McLaughlin, December 2001 -